ABSTRACT: Turbulent mixing is an integral aspect of aquatic ecosystems. Turbulence affects ecosystem features ranging from phytoplankton blooms at large scales through microscale interactions in the plankton. Enclosed experimental ecosystems, if they are to mimicthe function of natural ecosystems, also must mimic natural turbulence and its effects. Large-scale velocity gradients and unstable buoyancy fluxes generate turbulent mixing in nature, most often at the surface and bottom boundaries and in the pycnocline.Large eddy sizes are controlled by the mixed layer depth, boundary layer thickness, or overturning length in the pycnocline. Turbulent energy cascades through smaller and smaller eddy scales until it can be dissipated by molecular viscosity at thesmallest scales. In contrast, artificial apparatuses frequently are used to generate turbulent mixing in the interior of experimental ecosystem enclosures. Large eddy sizes are controlled by the size of the generation apparatus, and they usually are muchsmaller than in nature. Mismatched large eddy length scales and differences in turbulence generation mechanisms are responsible for the difficulties in mimicking natural turbulent mixing in experimental enclosures. The 2 most important turbulenceparameters to consider in experimental ecosystem research are overall mixing time, T_m, and turbulence dissipation rate, ε. If the levels, spatial distributions, and temporal variability of T_mand ε can be matched between an enclosure and the natural system it is to model, then potential mixing artifacts can be minimized. An important additional consideration is that benthic ecosystems depend on the time-averagedboundary layer flow as much as the turbulence. Existing designs for mixing experimental ecosystems are capable of reasonably representing some aspects of natural turbulent mixing. Paddle and grid stirring are the best available techniques for water columnmixing, and flumes are best for benthic turbulence. There is no design at present that represents both environments adequately. More work also is needed on mixing of flexible-wall in situ enclosures. A more serious problem, however, is thatturbulent mixing in experimental ecosystem studies too often is ignored, inadequately characterized, or unreported. Several methods are available for reasonable characterization of mixing in enclosures without sophisticated technology, and the technologyfor direct velocity measurements is becoming more accessible. Experimental ecosystem researchers should make a concerted effort to implement, characterize, and report on turbulent mixing in their enclosures.